Sign In

Communications of the ACM

ACM TechNews

Why the Card Game Hanabi Is Next Big Hurdle for AI

View as: Print Mobile App Share:
Hanabi presents a challenge for artificial intelligence because it requires communication and cooperation.

Alphabet Google Brain and DeepMind researchers hope to train artificial intelligence to master the card game Hanabi.

Credit: Matthias Balk/DPA/Zuma Press

Researchers at the Alphabet Google Brain project and DeepMind unit hope to train artificial intelligence (AI) to master Hanabi, a card game in which players cooperate with one another, with everyone winning or losing depending on how well they communicate through their play.

DeepMind's Nolan Bard said, "Cooperative games are different and hard because you need to be able to agree on a way that you're communicating with each other in order to play the game well, and doing that agreement requires collaboration among all the players."

The researchers think a Hanabi-playing AI could provide insights into more fluid human-machine interaction, in contexts like conversation and autonomous driving.

The project has thus far yielded Hanabi-playing programs that can win a lot, provided they play with other similarly programmed bots.

The researchers consider their biggest challenge to be facilitating "ad hoc" play, involving teammates whose playing styles are not previously known.

From The Wall Street Journal
View Full Article - May Require Paid Subscription


Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account